Neural network based high efficiency drive for interior permanent magnet synchronous motors compensating EMF constant variation

An optimum d-axis current makes na interior permanent magnet synchronous motor (IPMSM) drive on maximum efficiency. However, the optimum d-axis current cannot be computed in real-time because the calculation includes the cube root. In this paper, a neural network based high efficiency drive strategy for IPMSM is presented. The neural network is employed as an identifier to determine the optimum d-axis current. The neural network can determine the optimum d-axis current in real-time. Furthermore, the accuracy is hardly degraded irrespective of the variation of the EMF constant. Simulation results confirm the validity of the proposed approach